Introduction

Anemia and exposure to red blood cell (RBC) transfusion are predictors of adverse events after cardiac and non-cardiac surgery [1–9]. Differentiating the negative effects of these two factors is difficult since preoperative anemia is itself associated with increased need of RBC transfusions [7, 10–14]. On the other hand, preoperative anemia is relatively common in patients undergoing cardiac surgery [10] and is a result of multiple causes which, in turn, are risk factors for poor outcome in these patients [15]. The aim of the present study was to evaluate the independent effect of preoperative anemia on the outcome after coronary artery bypass grafting (CABG).

Materials and methods

Patients

The present study includes 2761 consecutive patients who underwent isolated CABG from June 2006 to December 2013 at the Oulu University Hospital, Finland. Complete data on pre-, intra- and postoperative variables were available in these patients. Data on the preoperative use of antithrombotics were retrospectively collected. Data on the types and amount of blood products such as RBCs, platelets and solvent/detergent-treated plasma (Octaplas; Octapharma AG, Lachen, Switzerland) were retrieved from a prospective electronic hospital registry. The amount of transfused blood products was estimated from the day of operation until discharge or up to a maximum of 30 days after the operation. Data on the amount of chest drainage output at 12 h after surgery were retrieved from a prospective electronic registry of the intensive care unit. Preoperative anemia was defined a hemoglobin level <12.0 g/dL in women and <13.0 g/dL in men [16]. Clinical variables were defined according to the EuroSCORE II definition criteria [17]. The E-CABG classification of bleeding severity [18] was employed to stratify the severity of perioperative bleeding based on the type and amount of blood products transfused and the need of resternotomy for excessive bleeding. The E-CABG bleeding grades [18] are: Grade 0, no transfusion of blood products or transfusion of one unit of RBCs; Grade 1, transfusion of platelets, fresh frozen plasma/Octaplas and/or 2–4 units of RBCs; Grade 2, transfusion of 5–10 units of RBCs and/or reoperation for bleeding; Grade 3, transfusion of >10 units of RBCs. Data on patients’ death were provided up to January 31, 2016 from the Finnish Population Registry Center, which collects the certificates of death of all inhabitants of Finland. We assume that there are no missing data on any immediate and late death of this study population.

Inclusion criteria

Patients who underwent any elective, urgent or emergency isolated CABG were included in this analysis.

Perioperative antithrombotic treatment

Aspirin was discontinued for 7 days until 2012 and later on was continued until surgery. Warfarin was discontinued for 2 days before surgery. Enoxaparin was used preoperatively only in patients with acute coronary syndrome. Clopidogrel, prasugrel and ticagrelor were discontinued for at least 5 days when feasible.

Heparin (3.0 mg/kg) was administered intraoperatively to maintain an activated coagulation time of longer than 450 s, and it was neutralized at the end of the procedure by protamine sulfate (3.0 mg/kg). Aprotinin was not used in any of these patients. Tranexamic acid was administered intraoperatively at discretion of the anesthesiologist. RBCs were transfused on the operation day if hemoglobin was less than 90 g/L. Later, RBCs were transfused if hemoglobin was less than 80 g/L. Octaplas as well as platelets were transfused according to the amount of intra- and postoperative bleeding, INR levels and platelet count.

Enoxaparin (40–80 mg once-a-day) was started on the evening of the operation day in those patients without excessive bleeding (<1000 mL). Aspirin 100 mg was restarted on the first postoperative day. Warfarin was restarted on the first postoperative day in patients on chronic oral anticoagulation unless excessive bleeding occurred. Warfarin was started de novo in case of persistent atrial fibrillation. Clopidogrel and ticagrelor were used postoperatively only in case of allergy to aspirin or recent percutaneous coronary intervention.

Operative techniques and management of chest drainages

Intermittent antegrade and retrograde cold blood cardioplegia was used during on-pump CABG. Epiaortic ultrasound was performed according to the surgeon’s preference. Octopus stabilizer (Medtronic, Minneapolis, MN) as well as intracoronary shunts were routinely used in patients who underwent off-pump surgery. All blood lost during the operation was collected into a cell saver reservoir and washed. Salvaged red blood cells were transfused during or at the completion of the operation. Mediastinal/pleural blood was collected after surgery in a sterile collection chamber connected to 15 cm H2O wall suction via an underwater seal and then discarded.

Outcomes

The primary outcomes of this study were 30-day and late mortality. Secondary outcomes were the length of intensive care unit stay, stroke, atrial fibrillation, ventricular fibrillation or asystole, permanent pacemaker implantation, postoperative use of antibiotics, deep sternal wound infection, mediastinitis, low cardiac output syndrome, repeat revascularization, surgery for gastrointestinal complications, acute kidney injury according to the Kidney Disease: Improving Global Outcomes (KDIGO) criteria based on peak creatinine level during the in-hospital stay [19], nadir hemoglobin level during the postoperative period and chest drain output 12 h after surgery. Low cardiac output syndrome was defined as postoperative cardiac index <2.0 L/min/m2 as measured at least twice. The other outcomes were defined according to the E-CABG definition criteria as previously reported [18].

Statistical analysis

Statistical analysis was performed using SPSS statistical software (version 23.0, IBM Corporation, New York, USA). No attempt to replace missing values was made. Continuous variables are reported as the mean and standard deviation. Nominal variables are reported as counts and percentages. Fisher’s exact test, Chi-square test and Mann–Whitney tests were used for univariate analysis. Risk estimates are reported as beta coefficients, odds ratio (OR) and hazard rate (HR) with 95% confidence interval (95% CI). Observational studies do not provide randomization, and therefore, propensity score matching method was employed to select two groups of patients with and without preoperative anemia having similar baseline and operative characteristics. The propensity score was estimated using a non-parsimonious logistic regression model with the preoperative anemia/non-anemia as the dependent variable. The following variables have been included as covariates: age, gender, body mass index, platelets count, eGFR, dialysis, pulmonary disease, diabetes, stroke, extracardiac arteriopathy, neurological dysfunction, atrial fibrillation, previous percutaneous coronary intervention, previous cardiac surgery, left ventricular ejection fraction ≤50%, recent myocardial infarction, critical preoperative status, preoperative intra-aortic balloon pump, recent ventricular arrhythmia, cardiac massage, urgency of the operation, off-pump surgery, epiaortic ultrasound, diseased ascending aorta, bilateral mammary artery grafts and radial artery graft. One-to-one propensity score matching was performed using a caliper width of 0.02 of the standard deviation of the logit of the propensity score. Analysis of the standardized differences after matching was used to assess the balance between the characteristics of propensity-matched pairs. The outcomes of matched groups were evaluated by the t test for paired sample for continuous variables and the McNemar test for dichotomous variables. Difference in overall survival at 9 years was evaluated by the Kaplan–Meier method. Logistic regression, linear regression, ordinal regression and Cox proportional hazards analyses of propensity-matched pairs were adjusted for the E-CABG bleeding grades in order to evaluate the impact of blood transfusion and resternotomy for excessive bleeding on the outcomes. In another regression model, preoperative anemia was adjusted for the number of transfused units of RBCs. All tests were two sided with the alpha level set at 0.05 for statistical significance.

Results

Baseline and operative data

In this series, 662 patients (24.0%, 22.8% of men and 28.6% of women) had preoperative anemia. Baseline and operative characteristics of these patients are summarized in Tables 1 and 2. The prevalence of comorbidities was significantly higher among anemic patients. In particular, anemic patients were significantly older, had a higher prevalence of recent myocardial infarction and were operated mostly on urgent/emergency basis. Operative factors were similar between the study groups.

Table 1 Baseline characteristics
Full size table
Table 2 Operative data
Full size table

Outcomes in the overall series

Outcomes in overall series are summarized in Tables 3 and 4. In-hospital, 30-day and long-term mortalities were significantly increased in patients exposed to anemia (P < 0.0001 in all). Unadjusted 8-year survival in non-anemic and anemic patients was 80.2 and 62.4%, respectively.

Table 3 Bleeding-related outcomes
Full size table
Table 4 Outcomes in the overall series and in propensity score matched cohorts
Full size table

Preoperative anemia was associated with a significantly increased risk of exposure to blood products (Table 3) and lower nadir hemoglobin values (P < 0.0001), whereas blood loss at 12 h was lower (484 ± 416 vs. 509 ± 388 mL, P = 0.029) when compared to non-anemic patients.

In univariate analysis, preoperative anemia was associated with significantly longer stay in the intensive care unit, atrial fibrillation, ventricular fibrillation or asystole, postoperative use of antibiotics and acute kidney injury (P < 0.0001 in all). The rates of postoperative stroke (P = 0.001), low cardiac output syndrome (P = 0.003) and surgery for gastrointestinal complications (P = 0.001) were also significantly higher among anemic patients.

Propensity score matched analysis

Logistic regression (Hosmer–Lemeshow’s test: P = 0.248) provided a propensity score with an area under the receiver-operating characteristics curve of 0.773 (95% CI 0.753–0.793). Propensity score matching (caliper width: 0.02) resulted in 560 patient pairs with similar baseline and operative covariates as confirmed by absolute standardized differences <10% in all baseline and operative covariates (Fig. 1).

Fig. 1
figure 1

Absolute standardized difference before and after propensity score matching in the cohorts of patients with and without anemia

Full size image

In these propensity-matched pairs, in-hospital (3.0% vs. 3.2%, P = 1.00) and 30-day mortality rates (3.9% vs. 3.9%, P = 1.00) were similar in non-anemics and anemics. However, Kaplan–Meier analysis showed a better survival among non-anemics at mid-term, but the difference somewhat decreased at long-term follow-up (at 8 years, 68.2% vs. 66.3%, P = 0.047, Table 4).

Exposure to RBC transfusion was significantly higher among anemic patients (Table 3). Similarly, nadir hemoglobin level was significantly lower among patients with preoperative anemia (mean 7.6 ± 0.9 vs. 8.3 ± 1.1 g/dL, P < 0.0001), but the amount of blood loss at 12 h (mean 477 ± 414 vs. 494 ± 376 mL, P = 0.408) and rates of reoperation for bleeding (5.2% vs. 7.1%, P = 0.222) were similar between the study groups.

In these propensity score matched pairs, preoperative anemia was associated with a significantly higher rate of acute kidney injury not requiring dialysis (26.3% vs. 15.9%, P < 0.0001), without any increased risk of other early adverse events.

In view of the increased need of RBC transfusion among patients with preoperative anemia (Table 3), analyses in these propensity score matched pairs were further adjusted for the severity of perioperative bleeding/amount of transfusion as stratified by the E-CABG bleeding classification. These adjusted analyses showed that preoperative anemia was associated only with an increased risk of acute kidney injury not requiring dialysis (OR 1.50, 95% CI 1.10–2.03), whereas other early outcomes were similar between anemics and non-anemics (Table 3).

Cox proportional hazards model in propensity-matched pairs including preoperative anemia and the severity of preoperative bleeding stratified by the E-CABG bleeding classification showed that anemia (HR 1.10, 95% CI 0.86–1.39, Fig. 2) was not an independent predictor of late death. Instead, this regression model showed that the E-CABG bleeding classification was an independent predictor of mortality (compared to grade 0: grade 1, HR 1.93, 95% CI 1.37–2.73, grade 2, HR 2.19, 95% CI 1.50–3.18, grade 3, HR 5.59, 95% CI 3.34–9.39, P < 0.0001). Similarly, anemia adjusted for the amount of transfused units of RBCs was not associated with poorer survival (HR 1.16, 95% CI 0.91–1.47), whereas the number of transfused RBCs was associated with increased risk of late death (per RBC unit: RR 1.06, 95% CI 1.04–1.08).

Fig. 2
figure 2

Cox proportional hazards estimate of survival in patients with and without anemia as adjusted by the E-CABG bleeding severity classification (HR, 1.10, 95% CI 0.86–1.39)

Full size image

Discussion

The main finding of this study is that preoperative anemia is not associated with poorer survival when adjusted for baseline covariates and severity of perioperative bleeding.

Several studies showed that RBC transfusion is associated with increased mortality and morbidity after cardiac surgery [8, 20, 21]. In turn, anemic patients are more likely to receive transfusions [7, 10–14] and preoperative anemia itself has been shown to be a risk factor for adverse events [4, 22–24]. However, the interaction between the anemia status and the increased need of transfusion makes difficult to disentangle the individual prognostic impact of these two factors. In their seminal contribution, Loor et al. [1] demonstrated the independent role of anemia and transfusion on the outcomes after cardiac surgery and concluded that both may have a significant and independent impact on adverse events. However, anemia and transfusion were assessed only intraoperatively, and this prevented conclusive results on this issue. Indeed, using a different methodology and evaluating anemia, blood loss and transfusion during the entire postoperative period, preoperative anemia was not a risk factor for late mortality in propensity score matched pairs adjusted for exposure to blood products. The only adverse event associated with preoperative anemia was an increased risk of acute kidney injury. These findings were observed although preoperative anemia was associated with significantly lower nadir hemoglobin levels without a difference in the amount of postoperative blood loss. However, some investigators consider anemia as a risk factor for increased bleeding because of its possible associated coagulopathy [25, 26].

Since preoperative anemia is a modifiable condition, the hypothetical possibility to reverse its related risk of adverse events is intriguing. Evidence from previous studies suggests that the prevalence of preoperative anemia might be higher in cardiac surgery [4, 5, 10, 22], than in the general population of similar age, from developed countries [27–29]. Increased comorbidities among patients requiring revascularization might explain the difference. Older age [22], lower baseline eGFR [19] and poor left ventricular ejection fraction [30] are associated with higher incidence of anemia. In addition to these factors, the present study identified recent myocardial infarction and critical preoperative status as significantly associated with preoperative anemia. We speculate that lower hemoglobin values observed in these two patient groups might be due, in some patients with critical conditions, to hemodilutional anemia secondary to substantial preoperative infusion of fluids.

However, the potential benefits of preoperative hemoglobin optimization are still unclear. Recent studies on the issue reported promising results. Preoperative treatment with EPO and/or intravenous iron is considered reasonable [31, 32] and shown to reduce the need for RBC transfusion [33–35]. Hemoglobin optimization > 12 g/dL was observed to improve the outcomes of Jehovah’s witnesses undergoing cardiac surgery [36].

Our study has some limitations which should be acknowledged. First, data collection was performed retrospectively. However, data on the amount of blood loss and use of blood products were recorded prospectively. This makes the estimation of perioperative bleeding and exposure to blood products rather reliable. Second, analysis limited to patients undergoing elective CABG would have enabled the exclusion of the impact of potential preoperative hemodilutional anemia, but it would have reduced the statistical power of this analysis. Importantly, propensity score matching provided pairs with similar prevalence of risk factors for critical prognostic importance. Third, this study does not address the possible multiple causes underlying preoperative anemia, and some of them might have a prognostic impact which is left unrecognized. The strengths of the study are a relatively large database and a long follow-up with complete survival data.

Conclusions

In this study, preoperative anemia in patients undergoing CABG was associated with significant baseline comorbidities. When adjusted for important baseline characteristics and operative factors as well as for the severity of perioperative bleeding and the amount of transfused blood products, anemia was not associated with increased risk of adverse events. Increased exposure to blood transfusion among anemic patients may be the determinant of their poorer late survival.